w2v-fine-tune-test-no-ws2

web2savar/w2v-fine-tune-test-no-ws2

w2v-fine-tune-test-no-ws2 is a automatic speech recognition model from web2savar. Use it when you need speech turned into text. It is set up for transformers.

should probably proofread and complete it, then remove this comment. --

Downloads · 30 days

2% of all-time downloads

All-time downloads

336

Public

Parameters

606M

43.6 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors2.4 GB · 100%

At a glance

Task: Automatic Speech Recognition
Library: transformers
Model type: wav2vec2-bert
Access: Public
Created: Jan 24, 2024
Updated: Jan 24, 2024
SHA: 5a92b765

Base models

ylacombe/w2v-bert-2.0

Datasets

common_voice_16_0

Model card

From the Hugging Face model README

w2v-fine-tune-test-no-ws2

This model is a fine-tuned version of ylacombe/w2v-bert-2.0 on the common_voice_16_0 dataset. It achieves the following results on the evaluation set:

Loss: 0.1513
Wer: 0.1109

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 5e-05
train_batch_size: 32
eval_batch_size: 8
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_steps: 500
num_epochs: 10
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
2.192	0.22	300	0.2797	0.2985
0.2226	0.44	600	0.2989	0.3491
0.1941	0.66	900	0.2558	0.2451
0.1659	0.88	1200	0.2320	0.2289
0.1332	1.1	1500	0.2063	0.1971
0.1129	1.31	1800	0.1873	0.2029
0.1044	1.53	2100	0.1765	0.1856
0.1026	1.75	2400	0.1719	0.1752
0.0982	1.97	2700	0.1927	0.2023
0.0769	2.19	3000	0.1776	0.1671
0.0715	2.41	3300	0.1626	0.1634
0.0695	2.63	3600	0.1666	0.1654
0.0612	2.85	3900	0.1760	0.1609
0.0614	3.07	4200	0.1645	0.1593
0.0476	3.29	4500	0.1685	0.1593
0.048	3.51	4800	0.1790	0.1583
0.0489	3.73	5100	0.1578	0.1535
0.0456	3.94	5400	0.1610	0.1617
0.041	4.16	5700	0.1559	0.1439
0.0367	4.38	6000	0.1536	0.1436
0.0321	4.6	6300	0.1591	0.1449
0.0349	4.82	6600	0.1616	0.1419
0.0308	5.04	6900	0.1501	0.1401
0.0233	5.26	7200	0.1588	0.1394
0.0253	5.48	7500	0.1633	0.1356
0.0254	5.7	7800	0.1522	0.1339
0.0245	5.92	8100	0.1598	0.1371
0.0189	6.14	8400	0.1497	0.1324
0.0174	6.36	8700	0.1487	0.1270
0.0178	6.57	9000	0.1397	0.1286
0.0173	6.79	9300	0.1495	0.1281
0.0178	7.01	9600	0.1462	0.1222
0.0124	7.23	9900	0.1516	0.1225
0.0121	7.45	10200	0.1554	0.1190
0.0128	7.67	10500	0.1453	0.1228
0.0113	7.89	10800	0.1468	0.1178
0.0086	8.11	11100	0.1556	0.1186
0.0085	8.33	11400	0.1507	0.1154
0.0073	8.55	11700	0.1494	0.1169
0.0079	8.77	12000	0.1507	0.1152
0.0089	8.98	12300	0.1456	0.1137
0.0062	9.2	12600	0.1518	0.1127
0.005	9.42	12900	0.1534	0.1115
0.005	9.64	13200	0.1514	0.1110
0.0048	9.86	13500	0.1513	0.1109

Framework versions

Transformers 4.37.0.dev0
Pytorch 2.1.2+cu121
Datasets 2.16.1
Tokenizers 0.15.1

Task: Automatic Speech Recognition
Library: transformers
Type: wav2vec2-bert
Created: Jan 24, 2024
Updated: Jan 24, 2024